Alibaba is raising $10.2bn and spending all of it on AI
THE SO WHAT
Alibaba is effectively turning a HK$80bn equity raise into AI capex, even after a 75% profit hit from similar spend—public markets are being asked to underwrite an infra land grab. If you compete with or sell to hyperscalers, assume AI infra budgets are not capped by current P&L and plan for more aggressive pricing and capacity moves.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIHarvey’s first in-house model for legal work is here
Vertical AI is starting to cut its dependence on frontier APIs—Harvey post-trained Tenet on Moonshot’s Kimi K3 to displace the very models its cap table is tied to. If you’re building in a regulated vertical, you now have to decide whether your moat is workflow, data, or owning the model itself.
Applied AI‘There's nothing better than being able to simplify’: Workday SVP tells us why effective AI usage can be a “superpower” for your business
When a major HCM vendor frames AI as simplification, not disruption, it’s signaling that the near-term value is in de-cluttering workflows rather than moonshot automation. If you’re buying into these suites, focus your pilots on a few high-friction processes where “less clicks” is a measurable win this quarter.
Applied AIA lawsuit says Oura’s sleep stages are AI guesses with a coin flip’s odds
If a court tests whether Oura’s AI-derived sleep stages are “measurement” or just probabilistic guesses, every consumer AI health claim is suddenly on the hook for evidentiary standards. If you ship AI-driven metrics in regulated-adjacent spaces, tighten your validation, disclosures, and marketing copy before plaintiffs’ attorneys do it for you.
Applied AI‘We are hitting a different chapter’: OpenAI leader warns of threat of ‘persistent’ AI cyber-attacks
When a major lab talks about “ongoing, persistent” AI-driven cyberattacks, it’s a warning that offense is scaling faster than most orgs’ defenses. Treat advanced models as both a tool and a threat—invest in red-teaming, anomaly detection, and staff training that assume attackers have LLM-level capabilities.